In conventional cellular wireless access networks, a cell is covered by a BTS (base station transceiver) and all mobile terminals communicate with the BTS directly. With the addition of relays, a multi-hop network, including the BTS, relay nodes and mobile terminals, is set up. With relay nodes involved, the coverage of a wireless access network is improved. In such a wireless access network, there may be multiple routes for communicating between a terminal and the network. For example, a terminal can communicate directly with the BTS or indirectly via one or more relay nodes.
An example of a two-hop scenario is shown in FIG. 1A. Shown is a base station 10 having a coverage area 12, and a relay 16 having a coverage area 18. There is a link 22 between the base station 10 and the relay 16 which is typically a high capacity and very reliable wireless link. Shown are a number of mobile terminals 14 communicating directly with the BTS 10, and a number of mobile terminals 20 communicating with the relay 16 which relays communications for these mobile terminals to and from the BTS 10.
In such a network, the fixed relay node 16 is added to improve the coverage in the edge of a cell. Since the relay 16 is a fixed node, the channel between the base station 10 and the relay 16 can be a high quality channel implemented using any one of many advanced channel technologies, such as MIMO, which can provide improved capacity. However, the quality of the channel between the relay node 16 and a mobile terminal 20 is typically lower and less stable due to the mobility of the mobile terminals 20 and due to height differences in the location of the relays 16 and the mobile terminals 20. Because of this, it is possible that data will accumulate in the relay 16 after having been transmitted over the reliable channel between the BTS 10 and the relay 16. This requires the relay node 16 to have a significant buffer capacity, particularly in the cases where a long delay bound is allowed and where there are a lot of mobile terminals 20 that are served by the relay 16. While the illustration shows the relay 16 extending the physical coverage of the cell, this may not always be the case. The relay 16 may provide service to an area within coverage area 12 to enhance service in that area for example to improve rate coverage within coverage area 12. In that case, a decision to communicate directly or indirectly via the relay can be made dynamically.
FIG. 1B shows an example of a three-hop scenario. In this example, there is an additional relay 30 shown communicating with another wireless terminal 32. There are three wireless hops to get from the BTS 10 through to the wireless terminal 32.
As indicated above, in such networks, the communication between the BTS 10 and the mobile terminal 20,32 may be possible via more than one route. The route can be selected dynamically by a L2 (layer 2) function.
Compared to wire-line networks, a concern on a wireless access network is the unreliability of wireless links. In order to provide comparable quality of wire-line networks, two re-transmission protocols are implemented to improve the reliability of a wireless access network. L1 HARQ (layer 1 hybrid automatic repeat request) is implemented in layer 1 and the Radio Link Protocol (RLP, layer 2 ARQ) is implemented in layer 2. The L1 HARQ is used to improve the quality of each hop individually. Each relay implements the L1 HARQ function. The L2 ARQ is used to improve the quality of a wireless end-to-end connection. Thus the L2 ARQ is implemented in the BTS and terminals but not in the relay nodes. This results in a simple relay node function, which can be beneficial for a low cost wireless access network.
L2 functionality typically includes multiplexing/scheduling functions (implemented by the so-called multiplex sub-layer) and flow control functions. If RLP flow control were running on the relay and the base station, two instances of RLP would be required on the wireless terminals to accommodate dynamic route selection. Conventional wireless terminals have only one such instance.
RLP is NACK based and run end-to-end, meaning that the receiver generates a NACK when it receives an out of sequence packet, but otherwise does nothing. L1 HARQ is run on a per-hop basis and is ACK based meaning that for each L1 attempt, the receiver responds to indicate success or failure of the attempt. Typically after some number of failed L1 attempts, the transmitter for that hop moves on to the next packet resulting in an out of sequence packet at the receiver which in turn causes a NACK to be generated.